Three broad goals underlie this continuation application for a multicenter collaborative study of the treatment of panic disorder: 1) completion of the originally proposed study, comparing cognitive behavioral panic control treatment (PCT), imipramine or placebo administered double blind (MED) and the combination (COM), 2) comparison of treatment durability over 6 or 24 month maintenance with further 24 month follow-up, and 3) cross-over treatment of MBD and PCT nonresponders to the opposite treatment. It is clearly important to complete enrollment of the remaining 37% of 480 patients needed to answer key questions related to the relative efficacy of PcT, MED and COM. Study enrollment began somewhat later than we predicted because of rigorous attention to state-of-the-art assessment and quality control procedures. Careful thought about essential assessment domains appears to have been well worth the effort. Not only are the procedures we developed working well, but they have been used already by other investigators in the field and have influenced recent consensus on assessment of panic disorder. We devoted great care to training and certifying therapists and to developing extensive quality assurance procedures, essential to a cross site study. PCT adherence ratings are among the most detailed and rigorous in the field, while our procedures for MED training, certification and adherence monitoring reflect to a level of quality control rarely undertaken in psychopharmacology trials. Study of long term maintenance and follow-up is a natural and important extension of the efficacy study. With efficacy of short term treatment for panic disorder well established, there is growing recognition of the need to address durability of treatment, especially medication. Naturalistic follow-up of patients who participated in efficacy trials document continued intermittent symptoms and ongoing functional impairment even with continued naturalistic treatment, and a high rate of relapse following medication discontinuation. However, there are no prospective studies of long term outcome using rigorous, blinded assessment procedures and quality controlled maintenance treatment. There is a need to determine course and outcome during and after optimal long term maintenance treatment. A second natural extension is to examine whether nonresponders to one treatment modality will respond to the other. Although this is an obvious question, few studies to date have investigated sequential treatment strategies. Such approaches could provide considerable useful information. Our original study will provide definitive information for clinicians regarding the best choices for acute treatment of panic disorder. We now recognize that the cohort of subjects we have assembled also provide a unique opportunity to conduct the first large scale study of treatment durability and the first systematic study of effectiveness of cross-over treatment for nonresponders. Thus, we are proposing to add these aims.